Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Jay Vosburgh <jay.vosburgh@canonical.com>
To:     =?us-ascii?Q?=3D=3FUTF-8=3FB=3F5a2Z5a6I6ZGr=3F=3D?= 
        <sunshouxin@chinatelecom.cn>
cc:     vfalico@gmail.com, andy@greyhouse.net, davem@davemloft.net,
        kuba@kernel.org, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org, huyd12@chinatelecom.cn
Subject: Re: [RESEND PATCH v5] net: bonding: Add support for IPV6 ns/na to balance-alb/balance-tlb mode
In-reply-to: <5c8a286b-932b-135f-5da4-68d18cfcb891@chinatelecom.cn>
References: <20220110090410.70176-1-sunshouxin@chinatelecom.cn> <11285.1641842636@famine> <5c8a286b-932b-135f-5da4-68d18cfcb891@chinatelecom.cn>
Comments: In-reply-to =?us-ascii?Q?=3D=3FUTF-8=3FB=3F5a2Z5a6I6ZGr=3F=3D?=
 <sunshouxin@chinatelecom.cn>
   message dated "Fri, 14 Jan 2022 17:07:04 +0800."
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Date:   Sun, 16 Jan 2022 18:02:11 -0800
Message-ID: <14104.1642384931@famine>
Precedence: bulk

=E5=AD=99=E5=AE=88=E9=91=AB <sunshouxin@chinatelecom.cn> wrote:
[...]
>> 	As for the RLB functionality (i.e., the balance-alb remote to
>> local load balance), that is not implemented for IPv6 and this patch is
>> not providing an implementation of the RLB logic for IPv6, so I'm
>> unclear why you expect it to work, or what the "mismatch Bond6
>> specification" is.
>>
>> 	To be clear, implementing RLB for IPv6 would include what this
>> patch is doing (adjusting the content of NS/NA datagrams), but a
>> complete implementation requires additional logic that isn't here, e.g.,
>> adding IPv6 logic to the RLB rebalance code, connecting NS/NA
>> manipulation to rlb_choose_channel(), and likely other things that don't
>> come immediately to mind.
>>
>> 	In summary, it sounds to me like the actual bug originally
>> reported (with the now-omitted diagram) would be resolved by assigning
>> NS/NA datagrams to the curr_active_slave, and supporting RLB for IPv6 is
>> a larger project than what's provided by this patch.  Am I understanding
>> correctly?
>
>
>Thanks your comment.
>For the simplify, I would like to resolve the inconsistent mac at first by
>assigning NS/NA datagrams to the curr_active_slave by V6 soon.
>Supporting RLB for IPv6, it looks like hard a bit and I wonder if we can
>resolve it in another patch=EF=BC=9F
>any comments?

	I'm in agreement that the first step should be solving the
immediate TLB NS/NA problem, and the larger task of implementing RLB for
IPv6 can be done separately.

	-J

---
	-Jay Vosburgh, jay.vosburgh@canonical.com